Research On The Improved Hybrid Beetle Genetic Algorithm In Function Optimization

The beetle antennae search algorithm(BAS)is a new type of meta-heuristic search algorithm,which relies on single-individual search characteristics to simplify the complexity of the algorithm and shorten its operation time.As a result,it has very good optimization speed and convergence accuracy in solving the problems with unimodal function characteristics.However,in solving the problems with multimodal characteristics,the expected convergence value can not be guaranteed because it is easy to be caught in the local optimal solution.Genetic algorithm(GA)is a kind of meta-heuristic algorithm with relatively perfect theory.How to maintain the balance of mining and exploration(i.e.,using the existing solution or exploring new solutions)in the process of evolution is the key problem of the genetic algorithm.This problem is directly related to the diversity of the population.Because of the controllable diversity of genetic algorithm,it has a great advantage in the optimization of multimodal problems.However,the diversity of the algorithm will lead to slow convergence in the optimization of the unimodal problems.In order to overcome the shortcomings of the above two kinds of algorithms and to utilize the respective advantages of the two algorithms,this paper presents an optimization algorithm that mixed the beetle antennae algorithm with the genetic algorithm.The hybrid algorithm improved both algorithms in their own strong point,and then mixed them to form a new algorithm.In short,the BAS algorithm is improved by further enhancing the fast convergence speed of the algorithm,and the GA is improved by increasing the richness of the population.The specific improvements of the hybrid algorithm are:(1)A new multi-directional individual exploration model is proposed,and based on this model,the strategy of probing feedback is proposed to update the position of the beetle individual,as a result,the optimization speed and the accuracy of the solution are further improved.(2)A disturbing operator which can make disturbance with small probability to the individual of the beetle is designed to reduce the risk of falling into the local optimal.(3)Using Tent chaotic sequence generator to produce the initial population,thus improving the coverage ratio of the population to the solution space.(4)A new crossover operator is proposed to improve the diversity of the population.(5)The intersection probability and mutation probability are adaptively chosen,so as to balance the population richness of the algorithm before and after operation.(6)The two improved algorithms are fused according to a certain strategy to form a hybrid algorithm.Finally,taking the function optimization problem as an example,the hybrid algoritm is compared with various algorithms proposed by other researhers,and the simulation results show that the proposed hybrid algorithm has a good performance in both search and convergence performance.